MAF1731 - Microstructure and Trading Systems

B.Eng Financial Engineering

I.F. ALEJANDRO MICHEL MARRÓN

Jun 2021 | Repository: Link

---

Lab 3

Behavioral Economics: Speculative Trading

---

Abstract

This laboratory has the particular purpose of developing tools for the analysis of the performance of trading activity, from the point of view of the account and from the point of view of the person who operates the account (trader). Therefore, this laboratory is divided into 4 parts, one shows the data used throughout the laboratory while the other 3 parts are each related to the purposes of the complete module. The divisions and specific purposes are as follows:

- Descriptive statistics

- Performance Attribution Measures

- Behavioral Finance

---

1. Introduction

---

Behavioral finance is the study of the influence of psychology on the behavior of investors or financial analysts. It also includes the subsequent effects on the markets. It focuses on the fact that investors are not always rational, have limits to their self-control, and are influenced by their own biases. Throughout this laboratory a focus in identifying disposition effect, an example of an irrational influence in trading choices is present. A process is followed to find it in historical trading data.

---

2. Install/Load Packages and Depedencies

---

Below you'll find available the packages and dependencies used in the making and running of the notebook.

import numpy as np
import pandas as pd
import datetime 
import time
import warnings
warnings.filterwarnings('ignore')

pd.set_option('display.float_format', '{:.4f}'.format)

from functions import f_pip_size
from data import data_historical, data_pips, data_pips0
from main import historical_data_2,historical_data_3,historical_data_4, capital_evolution, mad, be_fi
from visualizations import g_2, g_1, g_extra_1, g_extra_2, g_extra_3, g_extra_4, g_3
import pandas_datareader as web
import plotly.graph_objects as go
import plotly.express as px

---

2.2 Python Packages

---

In order to run this notebook, it is necessary to have installed and/or have the requirements.txt file with the following:

• pandas
• numpy
• jupyter
• datetime
• time
• warnings
• plotly
• pandas_dataread

---

2.3 Files Dependencies

---

The following are the file dependencies that are needed to run this notebook:

• files/historical_data.csv : Historical data from the operations ran in MetaTrader5.
• files/instruments_pips.csv : A pip is a standardized unit and is the smallest amount by which a currency quote can change. In this file some of the most frequently traded CFD's tick sizes are saved, these are used to calculate the pip size of the instruments.

---

3. Data Description

---

In this segment the data used to build the laboratory is shown and described.

A pip is a standardized unit and is the smallest amount by which a currency quote can change. In this file some of the most frequently traded instruments and their respective tick sizes are saved, these are used to calculate the pip size of the instruments.

data_pips

	TickSize
Instrument
AU200AUD	1.0000
BCOUSD	0.0100
CN50USD	1.0000
CORNUSD	0.0100
DE10YBEUR	0.0100
...	...
XLMUSD	0.0000
LTCUSD	0.0100
XMRUSD	0.0100
XRPUSD	0.0001
IOTUSD	0.0001

132 rows × 1 columns

The closed operation's reported history from MetaTrader5's account is shown, it includes the times in which the position opened and closed, the position, the type of operation, the volume, the stop loss and take profit, profit,open price and close price. A new column with the time from the beginning to the end of the operation is added.

data_historical.head(5)

	Time	Position	Symbol	Type	Volume	Price	S / L	T / P	Time.1	Price.1	Commission	Swap	Profit	Time3
0	2021-06-08 16:54:29	41566202	LNKUSD	buy	11.0000	23.1984	NaN	NaN	2021-06-09 15:26:18	24.4500	0	-14.7500	1376.8000	81109.0000
1	2021-06-08 16:59:03	41566725	BTCUSD	buy	5.0000	32549.5500	31000.0000	44000.0000	2021-06-09 15:26:09	35154.7000	0	-93.3200	13025.7500	80826.0000
2	2021-06-08 16:59:57	41566764	ETHUSD	buy	5.0000	2439.8500	2305.7400	2750.0000	2021-06-09 15:26:15	2555.4500	0	-70.3800	5780.0000	80778.0000
3	2021-06-08 17:00:05	41566788	LTCUSD	buy	5.0000	154.3100	NaN	NaN	2021-06-09 15:26:23	162.6700	0	-44.8200	4180.0000	80778.0000
4	2021-06-08 17:00:12	41566796	XLMUSD	buy	5.0000	0.3291	NaN	NaN	2021-06-09 15:26:19	0.3462	0	-9.7600	851.5000	80767.0000

Some of the names are fixed to make it more readable.

historical_data_2.head(5)

	Open Time	Position	Symbol	Type	Volume	Open Price	S / L	T / P	Close Time	Close Price	Commission	Swap	Profit	Tiempo	Pips	Pips Accumulated	Profit Accumulated	Profit GP
0	2021-06-08 16:54:29	41566202	LNKUSD	buy	11.0000	23.1984	NaN	NaN	2021-06-09 15:26:18	24.4500	0	-14.7500	1376.8000	81109.0000	1251.6360	1251.6360	1376.8000	100.0000
1	2021-06-08 16:59:03	41566725	BTCUSD	buy	5.0000	32549.5500	31000.0000	44000.0000	2021-06-09 15:26:09	35154.7000	0	-93.3200	13025.7500	80826.0000	260515.0000	261766.6360	14402.5500	90.4406
2	2021-06-08 16:59:57	41566764	ETHUSD	buy	5.0000	2439.8500	2305.7400	2750.0000	2021-06-09 15:26:15	2555.4500	0	-70.3800	5780.0000	80778.0000	11560.0000	273326.6360	20182.5500	28.6386
3	2021-06-08 17:00:05	41566788	LTCUSD	buy	5.0000	154.3100	NaN	NaN	2021-06-09 15:26:23	162.6700	0	-44.8200	4180.0000	80778.0000	836.0000	274162.6360	24362.5500	17.1575
4	2021-06-08 17:00:12	41566796	XLMUSD	buy	5.0000	0.3291	NaN	NaN	2021-06-09 15:26:19	0.3462	0	-9.7600	851.5000	80767.0000	1703.0000	275865.6360	25214.0500	3.3771

Three new columns are added to the previous DataFrame. One showing the pips of each operation, and the next ones showing the accumulated pips and profit.

historical_data_3.head(5)

	Open Time	Position	Symbol	Type	Volume	Open Price	S / L	T / P	Close Time	Close Price	Commission	Swap	Profit	Tiempo	Pips	Pips Accumulated	Profit Accumulated	Profit GP
0	2021-06-08 16:54:29	41566202	LNKUSD	buy	11.0000	23.1984	NaN	NaN	2021-06-09 15:26:18	24.4500	0	-14.7500	1376.8000	81109.0000	1251.6360	1251.6360	1376.8000	100.0000
1	2021-06-08 16:59:03	41566725	BTCUSD	buy	5.0000	32549.5500	31000.0000	44000.0000	2021-06-09 15:26:09	35154.7000	0	-93.3200	13025.7500	80826.0000	260515.0000	261766.6360	14402.5500	90.4406
2	2021-06-08 16:59:57	41566764	ETHUSD	buy	5.0000	2439.8500	2305.7400	2750.0000	2021-06-09 15:26:15	2555.4500	0	-70.3800	5780.0000	80778.0000	11560.0000	273326.6360	20182.5500	28.6386
3	2021-06-08 17:00:05	41566788	LTCUSD	buy	5.0000	154.3100	NaN	NaN	2021-06-09 15:26:23	162.6700	0	-44.8200	4180.0000	80778.0000	836.0000	274162.6360	24362.5500	17.1575
4	2021-06-08 17:00:12	41566796	XLMUSD	buy	5.0000	0.3291	NaN	NaN	2021-06-09 15:26:19	0.3462	0	-9.7600	851.5000	80767.0000	1703.0000	275865.6360	25214.0500	3.3771

Two graphs below show the profit aquired from each instrument through different operations. The first one shows the normal data while the second one normalizes it through a logarithmic conversion, in an attempt to make it easir to visualize.

g_extra_2(historical_data_3)

g_extra_3(historical_data_3)

---

4. Descriptive statistics

---

Descriptive statistics provide simple summaries about the sample and about the observations that have been made. Such summaries may be either quantitative, i.e. summary statistics, or visual, i.e. simple-to-understand graphs. These summaries may either form the basis of the initial description of the data as part of a more extensive statistical analysis, or they may be sufficient in and of themselves for a particular investigation.

Some measures that are commonly used to describe a data set are measures of central tendency and measures of variability or dispersion. Measures of central tendency include the mean, median and mode, while measures of variability include the standard deviation (or variance), the minimum and maximum values of the variables, kurtosis and skewness.

In this particular case we are measuring the amount of operations, wheter they were or not successful and how successful they were, as a median.

The function below obtains the multiplier number to express the price difference in pips. Which will be useful for the rest of the laboratory.

f_pip_size("BTCUSD", data_pips0)

124   100.0000
Name: TickSize, dtype: float64

Descriptive information from the operations is shown, the amount of total operations, how many were or were not successful, it's effectivness and median profit both in US Dollars and pips.

historical_data_4.get('df_1_tabla')

	Measurment	Value	Description
0	Ops Totales	17.0000	Operaciones Totales
1	Ganadoras	12.0000	Operaciones Ganadoras
2	Ganadoras_c	12.0000	Operaciones Ganadoras de Compra
3	Ganadoras_v	0.0000	Operaciones Ganadoras de Venta
4	Perdedoras	5.0000	Operaciones Perdedoras
5	Perdedoras_c	5.0000	Operaciones Perdedoras de Compra
6	Perdedoras_v	0.0000	Operaciones Perdedoras de Venta
7	Mediana(Profit)	851.5000	Mediana de Profit de Operaciones
8	Mediana(Pips)	836.0000	Mediana de Pips de Operaciones
9	r_efectividad	0.7059	Ganadoras Totales/Operaciones Totales
10	r_proporcion	2.4000	Ganadoras Totales/Perdedoras Totales
11	r_efectividad_c	0.7059	Ganadoras Compras/Operaciones Totales
12	r_efectividad_v	0.0000	Ganadoras Ventas/Operaciones Totales

The bar chart below shows the amounts from the table in a visually appealing way.

import plotly.graph_objects as go


g_extra_1(historical_data_4.get('df_1_tabla'))

From the instruments used in the laboratory, the effectiveness ratio of the operations carried out is shown. Ranked from the best ones to the worst ones.

historical_data_4.get('df_2_ranking')

	Symbol	Rank
1	ETHUSD	1.0000
4	LNKUSD	1.0000
5	LTCUSD	1.0000
7	XLMUSD	1.0000
8	XMRUSD	1.0000
0	BTCUSD	0.6667
9	XRPUSD	0.5000
2	EURUSD	0.0000
3	IOTUSD	0.0000
6	USDJPY	0.0000

The graph shows the operations previously mentioned in a pie chart format.

g_1(historical_data_4.get('df_2_ranking'))

---

5. Performance Attribution Measures

---

Performance attribution, or investment performance attribution is a set of techniques that performance analysts use to explain why a portfolio's performance differed from the benchmark. This difference between the portfolio return and the benchmark return is known as the active return.

In the DataFrame below, the following is shown:

• The first column, named 'Timestamp' contains the day-by-day dates during the period of trading.

• The second column is 'Profit' which is the profit per day for each day of all the contents in the period you traded.

• The third column of this DataFrame is 'Accumulated Profit' which is the daily accumulated profit of the account's capital. In particular, the column 'Accumulated Profit' represents the daily capital of the trading account, which before any trading position was closed, started with an amount of 100,000 USD.

capital_evolution

	Profit	Accumulated Profit
Close Time
2021-06-09	31489.0500	131489.0500
2021-06-10	28280.0000	159769.0500
2021-06-11	0.0000	159769.0500
2021-06-12	0.0000	159769.0500
2021-06-13	0.0000	159769.0500
2021-06-14	-0.1200	159768.9300
2021-06-15	10439.3600	170208.2900
2021-06-16	-1223.1000	168985.1900
2021-06-17	0.0000	168985.1900
2021-06-18	0.0000	168985.1900
2021-06-19	0.0000	168985.1900
2021-06-20	0.0000	168985.1900
2021-06-21	-711.0000	168274.1900

In the DataFrame below, "mad", 8 different metrics are shown, which come from two particular Performance Attribution topics:

Sharpe:

• Measures the average return over the risk-free rate per unit of risk incurred. That is, it is an indication of the expected return of the trader above the risk-free rate, discounting the risk incurred. The higher the value of the ratio, it means that the numerator was very large or that the denominator was very small. This, in financial terms, could be interpreted that a Sharpe is high when there is an excess return after the high risk-free rate, and / or, that the volatility of the returns of the trading strategy was low. Another simpler way to think about it is that the trader made much more return than the risk he incurred.

Drawdown / Drawup:

• The maximum capital gain / loss that occurred in the evolution of capital. This measure can be calculated, both for information temporarily measured in the same intervals or equidistant (eg daily returns) and also for information obtained at different intervals or not equidistant (eg: performance per operation). This is possible since what is sought to obtain information is the beginning date, the end date and the magnitude of the period where the accumulation of capital loss / gain turned out to be the highest within the period of time analyzed.

mad

	Metric	Measurment	Value	Description
0	Sharpe Original	Amount	126.8110	Sharpe Ratio Original Formula
1	Sharpe Actualizado	Amount	835.2190	Sharpe Ratio Adjusted Formula
2	DrawDown Capital	Initial Date	[2021-06-15 00:00:00]	Initial Date of Capital's DrawDown
3	DrawDown Capital	Last Date	[2021-06-21 00:00:00]	Last Date of Capital's DrawDown
4	DrawDown Capital	DrawDown Capital $	-1934.1000	Maximum floating loss recorded
5	DrawUp Capital	Initial Date	[2021-06-09 00:00:00]	Initial Date of Capital's DrawUp
6	DrawUp Capital	Last Date	[2021-06-15 00:00:00]	Last Date of Capital's DrawUp
7	DrawUp Capital	DrawUp Capital $	38719.2400	Maximum floating profit recorded

The bar chart below shows a comparison between the Updated and Original Sharp on the left side, and the DrawnDown and DrawnUp maximum floating amounts recorded.

g_extra_4(mad)

The line chart below shows the accumulated profit throughout the historic data's dates, starting on Jun 8th and finishing on Jun 21st of 2021. Both the DrawUp and DrawDown are noted with dotted lines.

g_2(capital_evolution)

---

6. Behavioral Finance

---

Behavioral economics (also, behavioural economics) studies the effects of psychological, cognitive, emotional, cultural and social factors on the decisions of individuals and institutions and how those decisions vary from those implied by classical economic theory.

Behavioral economics is primarily concerned with the bounds of rationality of economic agents. Behavioral models typically integrate insights from psychology, neuroscience and microeconomic theory. The study of behavioral economics includes how market decisions are made and the mechanisms that drive public choice.

Behavioral finance is the study of the influence of psychology on the behavior of investors or financial analysts. It also includes the subsequent effects on the markets. It focuses on the fact that investors are not always rational, have limits to their self-control, and are influenced by their own biases.

There are 2 approaches to address the existence of this bias,

1) Due to perception error

2) Due to valuation error.

In this lab we will use the 2nd option to build this last part.

be_fi.get('Ocurrencias')

{'Cantidad': 2,
 'Ocurrencia 1': {'Timestamp': '06/15/2021, 15:31:44',
  'Operaciones': {'Ganadora': {'Instrumento': 'XRPUSD',
    'Volumen': 15.0,
    'Sentido': 'buy',
    'Profit Ganadora': 211.51},
   'Perdedora': {'Instrumento': 'BTCUSD',
    'Volumen': 5.0,
    'Sentido': 'buy',
    'Profit Perdedora': -1403.1}},
  'ratio_cp_profit_acm': 0.020392312573413632,
  'ratio_cg_profit_acm': 0.0066806569539034355,
  'ratio_cp_cg': 6.633728901706775},
 'Ocurrencia 2': {'Timestamp': '06/15/2021, 15:31:46',
  'Operaciones': {'Ganadora': {'Instrumento': 'XLMUSD',
    'Volumen': 20.0,
    'Sentido': 'buy',
    'Profit Ganadora': 120.5},
   'Perdedora': {'Instrumento': 'BTCUSD',
    'Volumen': 5.0,
    'Sentido': 'buy',
    'Profit Perdedora': -1403.1}},
  'ratio_cp_profit_acm': 0.020392312573413632,
  'ratio_cg_profit_acm': 0.0018123418524931207,
  'ratio_cp_cg': 11.643983402489626}}

The disposition effect refers to investors’ reluctance to sell assets that have lost value and greater likelihood of selling assets that have made gains. In this particular case it was observed through the historical data. With two cases of strong pressence shown.

be_fi.get('Resultados')

	Ocurrencias	Status Quo %	Aversion Perdida %	Sensibilidad Decreciente
0	2	16.6667	16.6667	No

g_3()

---

6. References

---

[1] Munnoz, 2020. Python project template. https://github.com/iffranciscome/python-project. (2021).

[2] Unkown. (2015). What is Behavioral Finance?. 06/29/2020, de CFI Sitio web: https://corporatefinanceinstitute.com/resources/knowledge/trading-investing/behavioral-finance/#:~:text=Behavioral%20finance%20is%20the%20study,influenced%20by%20their%20own%20biases.

[3] Shefrin, H., & Statman, M. (1985). The disposition to sell winners too early and ride losers too long: Theory and evidence. The Journal of Finance, 40, 777-790.